During courtship, many animals, including insects, birds, fish, and mammals, utilize acoustic signals to transmit information about species identity. Although auditory communication is crucial across phyla, the neuronal and physiologic processes are poorly understood. Sound-evoked chaining behavior, a display of homosexual courtship behavior in Drosophila males, has long been used as an excellent model for analyzing auditory behavior responses, outcomes of acoustic perception and higher-order brain functions. Here we developed a new method, termed ChaIN (Chain Index Numerator), in which we use a computer-based auto detection system for chaining behavior. The ChaIN system can systematically detect the chaining behavior induced by a series of modified courtship song playbacks. Two evolutionarily related Drosophila species, Drosophila melanogaster and Drosophila simulans, exhibited dramatic selective increases in chaining behavior when exposed to specific auditory cues, suggesting that auditory discrimination processes are involved in the acceleration of chaining behavior. Prolonged monotonous pulse sounds containing courtship song components also induced high intense chaining behavior. Interestingly, the chaining behavior was gradually suppressed over time when song playback continued. This behavioral change is likely to be a plastic behavior and not a simple sensory adaptation or fatigue, because the suppression was released by applying a different pulse pattern. This behavioral plasticity is not a form of habituation because different modality stimuli did not recover the behavioral suppression. Intriguingly, this plastic behavior partially depended on the cAMP signaling pathway controlled by the rutabaga adenylyl cyclase gene that is important for learning and memory. Taken together, this study demonstrates the selectivity and behavioral kinetics of the sound-induced interacting behavior of Drosophila males, and provides a basis for the systematic analysis of genes and neural circuits underlying complex acoustic behavior.

Indoor localization is an important technology for providing various location-based services to smartphones. Among the various indoor localization technologies, pedestrian dead reckoning using inertial measurement units is a simple and highly practical solution for indoor localization. In this study, we propose a smartphone-based indoor localization system using pedestrian dead reckoning. To create a deep learning model for estimating the moving speed, accelerometer data and GPS values were used as input data and data labels, respectively. This is a practical solution compared with conventional indoor localization mechanisms using deep learning. We improved the positioning accuracy via data preprocessing, data augmentation, deep learning modeling, and correction of heading direction. In a horseshoe-shaped indoor building of 240 m in length, the experimental results show a distance error of approximately 3 to 5 m.

Indoor pedestrian localization has been the subject of a great deal of recent research. Various studies have employed pedestrian dead reckoning, which determines pedestrian positions by transforming data collected through sensors into pedestrian gait information. Although several studies have recently applied deep learning to moving object distance estimations using naturally collected everyday life data, this data collection approach requires a long time, resulting in a lack of data for specific labels or a significant data imbalance problem for specific labels. In this study, to compensate for the problems of the existing PDR, a method based on transfer learning and data augmentation is proposed for estimating moving object distances for pedestrians. Consistent high-performance moving object distance estimation is achieved using only a small training dataset, and the problem of the concentration of training data only on labels within a certain range is solved using window warping and scaling methods. The training dataset consists of the three-axes values of the accelerometer sensor and the pedestrian’s movement speed calculated based on GPS coordinates. All data and GPS coordinates are collected through the smartphone. A performance evaluation of the proposed moving pedestrian distance estimation system shows a high distance error performance of 3.59 m with only approximately 17% training data compared to other moving object distance estimation techniques.